Planned switchover before a maintenance window
DBA · mid-market payments platform · 4 Patroni clusters
Situation
Friday evening change: move primary for storage work. Historically — patronictl over SSH, lag checked in three dashboards, confirm typed into chat. One missed pending_restart nearly shipped a surprise restart into the window.
In Concord
Open the project board, confirm leader and lag, read Confidence on Ops, run Safe Ops switchover with preflight and typed confirm. Audit row exported into the change ticket.
Outcome
Window started with a single evidence trail. No curl under pressure. Community covered three clusters; the fourth estate pushed the team to evaluate Pro.
Lost leader during peak — need the picture now
SRE · retail · mixed VM + CNPG
Situation
Alert: no leader on a critical shard. Grafana shows lag spikes; REST is only reachable via bastion. The runbook says “check DCS and VIP” — five tabs, two people on a bridge.
In Concord
SSH profile already saved. Topology and Trust probes (DCS last-seen, write VIP TCP) in one view. Timeline correlates alerts with recent ops. Pause considered only after Confidence leaves blocked.
Outcome
Bridge shared a screenshot of evidence, not a paste of curl. Failover remained a Pro-gated, type-to-confirm path — not a panic click.
Partner estate with no path to the commercial API
Platform lead · regulated bank · air-gapped ops VLAN
Situation
Mac admin laptops cannot reach Stripe or api.goconcord.app. Procurement still wants signed entitlements and a paper trail for Pro features (K8s connect, longer audit retention).
In Concord
Community install from an internal mirror. Offline license JSON issued on the partner path, paste-activated in Settings → License. Docs for offline activate used as the operator runbook.
Outcome
Pro unlocked without opening the VLAN. Same Safe Ops UX as online customers; commercial events stay on the invoice / dual-control side, not in cluster telemetry.